![]() For this blog post, I decided that KEDA should watch my application’s queue length, which is the number of HTTP requests queued up in the ingress controller, not yet forwarded to the backend application. The HAProxy Kubernetes Ingress Controller exposes more than 100 different Prometheus-style metrics, which KEDA can watch and use to trigger autoscaling. In this blog post, you will learn how to use the open-source tool KEDA (Kubernetes Event-Driven Autoscaler) to monitor metrics coming out of the HAProxy Kubernetes Ingress Controller instead. The only shortcoming is that the built-in autoscaler, which is called the Horizontal Pod Autoscaler, can only monitor a pod’s CPU usage. Kubernetes ships with autoscaling baked in, giving you the power to scale out when the system detects an increase in traffic-automatically! As the volume of traffic to your application increases, you can create more application containers on the fly to handle it, in almost no time at all. One of the greatest strengths of containers is the ability to spin more of them up quickly. Since it was published, a new feature has been added to the ingress controller that lets you autoscale the number of ingress controller pods too! Read about it here. ![]() This blog post describes how to implement autoscaling of your application pods using KEDA and the HAProxy Kubernetes Ingress Controller.
0 Comments
Leave a Reply. |